Icaronycteris gunnelli, Rietbergen & van den Hoek Ostende & Aase & Jones & Medeiros & Simmons, 2023

Rietbergen, Tim B, van den Hoek Ostende, Lars W, Aase, Arvid, Jones, Matthew F, Medeiros, Edward D & Simmons, Nancy B, 2023, The oldest known bat skeletons and their implications for Eocene chiropteran diversification, PloS one 18 (4), pp. 1-20 : 3-12

Icaronycteris gunnelli sp. nov.

Holotype

AMNH: FM: 145747A,B (part and counterpart), an articulated skeleton including skull and mandibles with restoration ( Figs 1 and 2 ) .

Paratype

ROM:Palaeobiology-Vertebrate Fossils: 52666 , an articulated skeleton including skull and mandibles ( Fig 3 ) .

Locality, horizon, and age

The holotype was found by Terry Rickords on August 16, 2017 at the American Fossil Quarry (previously known as the Thompson Ranch South Quarry or (South) Dempsey Quarry), Lincoln County, Wyoming (N41˚51.872’ W 110˚40.491’), ± 1, 15 meter below the Tri-ash layer in the sandwich beds of the Fossil Butte Member ( FBM) ( Figs 4 and 5 ), Green River Formation ( GRF), late early Eocene (Lostcabinian, Wasatchian Biochron Wa-7). This locality preserves near-shore lacustrine deposits of Fossil Lake [ 40, 50]. The volcanic ash layer near the top of the FBM is radiometrically dated at 51.98 ± 0.34 Ma [ 51, 52]. Deposition rates are unknown in Fossil Lake. Deposition near-shore is more rapid than mid-lake [ 53], confounding efforts to estimate rates of deposition. Until reliable deposition rates are established there is no reliable means to estimate ages of specimens. The paratype was found in the same quarry as the holotype by Robert Kronner in 1994 in the bottom 60 cm of the sandwich beds of the FBM, Green River Formation.

Etymology

This species name is in honor of Gregg Gunnell in recognition of his extensive contributions to the understanding of fossil bats and chiropteran evolution.

Diagnosis

Icaronycteris gunnelli is the smallest known chiropteran from the Green River Formation (S1 Table) and is distinguished from other Eocene bats by the following combination of traits: claw present on wing digit I and II; tiny ossified third phalanx present on wing digit III–V; relatively short forearm and broad wing; relatively short, robust hind limbs with a possibly sutured tibia and fibula; lower canine tall and lanceolate; upper canine with anteromedial groove; diastema present between C and P1; P3 with raised lingual cingulum and a crown length relatively shorter than that of P4; P4 with well-developed lingual cingular cusp and labial cingulum; P4 length and width subequal; upper molars with strongly developed ectocingulum.

Description and comparisons

The skull of the holotype of Icaronycteris gunnelli is dorsoventrally crushed, so many cranial features, including structures of the basicranium, cannot be distinguished. However, the shape of the skull and several important traits of the skull and postcranium can be evaluated. Morphological traits of I. gunnellii are similar to those seen in Icaronycteris index and Onychonycteris finneyi except if explicitly stated otherwise below. The proclivous premaxilla in Icaronycteris gunnelli extends anteriorly beyond the level of the canine root and the nasal process of the premaxilla is well developed. The zygomatic arch appears complete and is broader than the mastoid region. Parietals appear to have not been inflated based on their rounded shape. The angular process of the dentary in I. gunnelli projects below the level of the occlusal plane, and the coronoid process is approximately two times the height of the condylar process.

The dental formula of I. gunnelli is I2 /3, C1/1, P3/3, M3/3 = 38. The upper incisors are orthodont and orthoclivous, and in this they resemble those of Onychonycteris finneyi . In contrast, the upper incisors in I. index are proodont and orthoclivous. The crowns of the upper incisors are not clearly differentiated from the root shafts in I. gunnelli , a condition similar to I. index but different from that seen in O. finneyi , in which the crown and root shaft of the upper incisors are clearly differentiated. The occlusal margin of I1 includes a large main cusp that is offset mesially from the axis of the tooth and tapers to a blunt point. A distal accessory cusp is present lateral to the main cusp on I1, with the accessory cusp approximately ¼ the size of the main cusp. The two upper incisors are subequal in height. Crown morphology of I2 is well developed and lingual cingulum is absent. Lack of lingual cingulum of the I2 is a characteristic also seen in I. index , whereas a lingual cingulum is present on this tooth in O. finneyi . I1 and I2 are subequal in height in I. gunnelli , also a condition shared with I. index . In contrast, in O. finneyi the height of I1 is less than that of I2. The upper canine (C) is separated from I2 by a diastema in all of these taxa.

The upper canine in I. gunnelli is elongated anteroposteriorly but does not project anteriorly. This tooth is not labially swollen and it lacks a raised cingulum or accessory cusp. No lingual cingulum is present, which is another trait shared with I. index . By contrast, O. finneyi exhibits a faint lingual cingulum on the upper C. The distal cingulum of the upper canine is small in I. gunnelli . A labial cingulum on the canine was not visible on the X-ray, but we could not determine if this indicated true absence or a lack of resolution in our images. An anteromedial groove is present on the upper canine, a structure that is not present in I. index or O. finneyi . A posteromedial ridge is present and the posterolingual and posterolabial surfaces of the canine are flattened. The upper canine has a tiny posterolateral accessory cusp, a structure that is absent in O. finneyi , and present but slightly smaller in I. index . The shape of this cusp is conical and is not clearly separated from the main cusp; it originates on the posterior face of the main cusp. A diastema is present between upper C and P1.

The upper premolars lie in line with the molar toothrow and no teeth are offset medially or laterally. The P1 is single rooted. This tooth has a well-developed crown which lacks a labial cingulum, similar to I. index but differing from O. finneyi , which has a labial cingulum on P1. The P 3 in I. gunnelli also lacks a labial cingulum, again as in I. index , with both Icaronycteris species differing from O. finneyi , which exhibits a weakly developed labial cingulum on P3. The lingual cingulum of P3 is weakly developed and lacks a cusp, but the edge of the cingulum is slightly raised. This differs from the condition in I. index and O. finneyi , both of which have a lingual cusp on P3. The crown length of P1 is shorter than that of P3. This condition is shared with I. index , while P1 and P3 are subequal in crown length in O. finneyi . The postparacrista of P3 extends as a single crest to the distal edge of the tooth and lacks a distal accessory cusp. Both P3 and P4 have three roots, but the crown length of P3 is shorter than that of P4. The latter trait contrasts with the condition in I. index and O. finneyi , both of which have a P3 that is longer than P4. The height of P3 is subequal to that of P 4 in I. gunnelli . This differs from both I. index , which has a P3 that is taller than P4, and from O. finneyi , in which P3 is shorter in height than the P4. No diastema present between the P3 and P4. The lingual cingulum of the P4 is very large and resembles that of I. index . In both of these taxa, the lingual cingulum forms a distinct lobe that extends posteriorly as far as the protocone of the first molar. This condition differs from that seen in O. finneyi , where the lobe of the lingual cingulum does not extend as far as the protocone of the first molar. The labial cingulum of the P4 is well developed in I. gunnelli and O. finneyi , but absent in I. index . The P 4 in I. gunnelli has three roots. The lingual cingulum of the P4 is very large and resembles that of I. index . In both of these taxa, the lingual cingulum forms a distinct lobe that extends posteriorly as far as the protocone of the first molar. This condition differs from that seen in O. finneyi , where the lobe of the lingual cingulum does not extend as far as the protocone of the first molar. The labial cingulum of the P4 is well developed in I. gunnelli and O. finneyi , but absent in I. index . A lingual cingular cusp is well developed on P 4 in I. gunnelli . This differs from the condition in I. index , in which the lingual cingular cusp is weakly developed. The condition of a lingual cingular cusp in O. finneyi cannot be determined from the specimens available. The P4 postparacrista is a single crest that extends as a continuous crest to the distal aspect of the tooth. The length and width of the P4 are subequal in I. gunnelli , in contrast to the conditions seen in I. index (P4 wider than long) and O. finneyi (P4 longer than wide).

The upper molars of I. gunnelli are characterized by a strongly developed ectocingulum. This is similar to the condition seen in O. finneyi , but differs from that in I. index , in which the ectocingulum is present but weak. A lingual cingulum and stylar shelf are present on both the M1 and M2, and the protocone and paracone are subequal in height. The protocone on both M1 and M2 is sharp and inclined anteriorly. The angle between the postparacrista and premetacrista is acute. The postparacrista contacts the premetacrista on the labial aspect of the tooth and closes the trigon basin labially on both M1 and M2. The trigon basin on both teeth is wide and shallow, and a mesostyle and mesostylar crest are absent. The contact between the postparacrista and premetacrista is more lingually displaced than the condition observed in I. index and similar to that of O. finneyi , and both M1 and M2 have a cuspule on the premetacrista as well as a small paraconule. The maximum mesiodistal length of M1 and M2 is less than half the distance between the paracone and metacone of the respective teeth.

The M1 has a long, straight, crestlike, parastyle that is not separated from the preparacrista. A single deep ectoflexus is present labial to the mesostyle. The preparacrista of M1 is subequal in length to the postparacrista, and the postmetacrista is about 1.5 times longer than the premetacrista. The trigon basin does not show any striations or enamel folds, instead being relatively smooth. The postprotocrista is oriented distolabially towards the metacone, and it extends labially to become confluent with the metacingulum and the edge of the hypocone shelf. A metaconule is present on the M1, but an endoloph is absent. The hypocone shelf is small and narrow but clearly distinct from the trigon basin. The M2 parastyle is crestlike and curved. Like M1, the M2 has a single deep exoflexus labial to the mesostyle. The metacone is located directly posterior to the paracone. The M2 postmetacrista, which is oriented distolabially towards the metacone, is 1.75 times longer than the premetacrista. An endoloph is present on M1, but is lacking on M2. The hypocone shelf on the M2 is very small and appears only as a narrow cingulum; no hypocone cusp is present.

The M3 is moderately reduced with a surface area <75% that of M2. The anterior segment of the M3 is nearly as large as that of the M2, and a long, crested curved parastyle and a paraconule are present. The premetacrista and postparacrista of M3 are subequal in length, and the metacone is present as a distinct cusp. A metastylar fovea is lacking on the M3, but the protocone and small hypocone shelf are present.

The three lower incisors in I. gunnelli are trilobed and somewhat procumbent; they are all similar in size and shape. The alveoli are evenly spaced, and there is no diastema between the last incisor and lower canine. The lower canine is tall, lanceolate, and lacks an anterior cuspule. The posterior cingulid on the canine is broad and a posterobasal cusp is strongly developed. The labial cingulum and cingular structures of the canine are not clearly visible given preservation of the holotype of I. gunnelli and hence cannot be evaluated.

As in other Eocene bats, I. gunnelli has three lower premolars. The p1 is large, single-rooted, cuspidate and has a distal cuspule. The crown length of p1 is less than that of p3. This is similar to the condition seen in I. index , but different from that of O. finneyi , in which p1 crown length is greater than p3 length. The p3 is premolariform, i.e., the cusps are not arranged in a tribosphenic pattern. Anterolingual and distal cuspules are present, but there is no lingual cuspule. Both p3 and p4 have two roots that are oriented longitudinally with respect to the long axis of the tooth row. The crowns of p3 and p4 are subequal in height and length but differ in morphology; in contrast to p3, the p4 is tribosphenic with a distinct trigonid and talonid. The paraconid of p4 consists of a single large cusp supporting a paracristid; this tooth also has a large metaconid, a long protocristid, and a long talonid. The protoconid is tall, sharp and slender. Labial structures on the premolars are not visible and therefore cannot be described in detail.

The lower molars of I. gunnelli lack a lingual cingulid and the trigonid fovea is open lingually. The hypoconulid is located on the distolingual border of the talonid, well separated from the entoconid and closer to the midline than to the entoconid. The height of the lingual cusps appears relatively low compared to the labial cusps. The entocristid is disrupted (the entocristid is short so that talonid basin drains lingually) on all three lower molars. The m1 and m2 are subequal in size and have similar crown morphology including a well-developed protoconid, paraconid, and metaconid. I. gunnelli and I. index both have a large hypoconulid on the m1 and m2, but in O. finneyi the hypoconulids on these teeth are small and low. The protoconid on m1 is markedly higher than the hypoconid, but is subequal in height to the protoconids on the m2 and m3. The m3 has two roots and bears a large hypoconulid.

The axial skeleton of I. gunnelli generally resembles that of other Eocene bats. An accurate vertebral count in the thoracolumbar region could not be determined due to poor preservation. The tail has 11 free caudal vertebrae. The scapula has an acromion process that extends anterolaterally beyond the glenoid fossa, and the lateral and dorsal edges of the acromion are flat. The infraspinous fossa is narrow and triangular shaped, tapering anteriorly towards the glenoid. A thick lip is present along the axillary border of the scapula. The coracoid process is short, stout, and appears to curve ventrolaterally.

Like other known Eocene bats, I. gunnelli has a well-developed wing skeleton with elongated hand and finger bones. The relative length of the humerus compared to the radius is ~76%, which is comparable to that of O. finneyi (~76%) while somewhat greater than that of I. index (~71%). The head of the humerus is spherical, and the distal articular surface of the humerus is laterally displaced. The distal end of the radius is wider than the mean width of the shaft. An ossified ulnar patella is present. The ulna and radius are fused distally starting at approximately the midpoint of the forearm. The metacarpal formula is IV ≥ V = III> II> I.

The hindlimb of I. gunnelli slightly differs from other Green River bats. The greater trochanter does not extend proximally as far as the femoral head, which differs from the condition seen in I. index and O. finneyi , in which the greater trochanter extends as far proximally as the level of the femoral head. The lesser trochanter is large and forms an extended flange in I. gunnelli as in O. finneyi . By contrast, the lesser trochanter is present only as a small tubercle in I. index . A third trochanter is absent in all three Green River bat taxa. The femur shaft is straight in anterior view and a femoral neck is absent. A small articulation is present between the distal ends of the tibia and fibula. The fibula is fully ossified and the proximal ends of the fibula and tibia are possibly sutured.

The first digit of the foot is shorter than the remaining digits. There is no visible contact between metatarsals of adjacent digits although intermetatarsal contact and/or facets may have been present, but are not detectable in our specimens. The shape of the proximal margin of the proximal phalanx seems to be flat in lateral view, but it is possible that this is due to taphonomic alteration. Foot digits II–V each have three phalanges.

Phylogenetic relationships

Phylogenetic analyses including all taxa resulted in three most parsimonious trees (MPTs) of 1511 steps ( Fig 6A ). The strict consensus tree recovered a sister relationship between Icaronycteris gunnelli and Icaronycteris index with high bootstrap support (bootstrap value 100). Onychonycteris finneyi fell sister to the Icaronycteris clade with lower support (bootstrap value 52), while Archaeonycteris trigonodon was recovered sister to the clade of Green River bats with low support (bootstrap value 37). Icaronycteris menui fell sister to the above clade in the strict consensus tree, also with low bootstap support (bootstrap value 46). The only other clade recovered in the strict consensus tree is the sister relationship between Rhinolophus ferrumequinum and Rhinopoma hardwickii , which received relatively high support (bootstrap value 76). Tachypteron franzeni in this analysis occurred in a polytomy with the previously mentioned clades, the remaining extant bats, Palaeochiropteryx tupaiodon , Hassianycteris messelensis , Icaronycteris ? menui , and I. sigei .

Due to the fragmentary nature of both Icaronycteris menui and I. sigei , we conducted a second analysis omitting those taxa. This analysis resulted in two MPTs of 1420 steps each ( Fig 6B ). The strict consensus tree again recovers an I. gunnelli and I. index sister relationship, with O. finneyi basal to that clade. Archaeonycteris trigonodon falls in a polytomy to the Green River bats in this analysis, along with a clade consisting of Palaeochiropteryx tupaiodon and Hassianycteris messelensis . The I. gunnelli + I. index clade again received high bootstrap support (bootstrap value 100), while the position of O. finneyi basal to that clade also received high support (bootstrap value 89). The position of A. trigonodon basal to the Green River clade received lower support (bootstrap value 54), as did the polytomy of the Green River bats + Archaeonycteris plus P. tupaiodon and H. messelensis (bootstrap value 56). Tachypteron franzeni in this analysis fell sister to a clade consisting of the extant emballonurid Saccopteryx bilineata and the extant phyllostomid Macrotus waterhousii with moderate support (bootstrap value 72).

Figures

Fig 1. Skeleton of Holotype of Icaronycteris gunnelli (FM.145747A) A) Dorsal view; B) Counterpart (FM.145747B). https://doi.org/10.1371/journal.pone.0283505.g001

Fig 2. Illustrated dentitions of Green River bats. A) Onychonycteris finneyi (ROM 55351A-B) upper P1-M3; B) Icaronycteris index (YPM-PU 18150) upper P1-M3; C) Icaronycteris gunnelli (FM.145747A) upper P1-M3; D) Icaronycteris gunnelli (FM.145747A) lower c1-m3, with reconstructed i1-i3, in labial view.Abbreviations of features discussed in the text: dc—distal cingulid; ec—ectocingulum; ef—ectoflexus; hs—hypocone shelf; lc—lingual cingulum; pbc—posterobasal cusp; prm—premetacrista; prp—preparacrista; ps—parastyle; psm—postmetacrista; psp— postparacrista; tal—talonid. https://doi.org/10.1371/journal.pone.0283505.g002

Fig 3. Skeleton of paratype of Icaronycteris gunnelli (ROM:Palaeobiology-Vertebrate Fossils:52666). https://doi.org/10.1371/journal.pone.0283505.g003

Fig 4. Remaining deposits of Fossil Lake sediments of the Green River Formation are illustrated in gray with highly-fossiliferous, deep-water, laminated limestone deposits of Eocene Fossil Lake in southwest Wyoming and poorly-fossiliferous, shallow-water, non-laminated limestones in northeast Utah and southeast Idaho. Eleven of more than 20 fossil quarries are still active today. Three active quarries (A, B and D) and one inactive quarry (C) have produced fossil bats currently held in public institutions. All other historic and active quarry locations are marked by small black dots. American Fossil Quarry (A) yielded I. gunnelli (AMNH.FM.145747 and ROM.52666), I. index (AMNH.FM.125000 and AMNH.FM.144215), I. cf index (WDC-CGR-115) and O. finneyi (AMNH.FM.142467 and ROM.55351). Thompson Ranch north quarry (B) yielded I. index (FMNH.PM.62096 and HMNS.PV.001468). The Holland brother’s quarry (C) was active for only one season in the mid-1930s and yielded I. index (YPM-PU.18150). The Smith Hollow quarry (D) yielded a poorly preserved and unidentified specimen (FOBU13777). Most bat specimens were recovered from two quarries operating continuously since the 1980s in nearshore deposits. A combination of proximity to the eastern shore and a large volume of rock excavated is the most probable explanation for the greater number of bat fossils being discovered in those locations. https://doi.org/10.1371/journal.pone.0283505.g004

Fig 5. A composite stratigraphic column of the Green River Formation showing the relative position of the Fossil Butte Member and stratigraphic levels where various bat fossils have been found. The Fossil Butte Member is the thinnest of the members in the Green River Formation, but was deposited when the lake was deepest, providing conditions that were optimal for exceptional preservation. Five units in the Fossil Butte Member (Sandwich Beds, 18-inch Layer, Minifish Bed, Gastropod Beds, and Upper Split-fish Bed) are actively quarried on private and stateowned land by companies collecting the abundant fossil fish to sell. Among the fish, one bat is found on average every two years. To date, all bats have come from the 18-inch Layer and Sandwich Beds. Bat specimens are coded by color: Icaronycteris gunnelli in red, I. index in orange, Onychonycteris finneyi in blue, and unidentified or indeterminate are black.More than 10 bat specimens are held in private collections (not shown here), but none are confirmed lower in the section than the two specimens of I. gunnelli. Left two columns modified after Buchheim et al, 2011 [40]. https://doi.org/10.1371/journal.pone.0283505.g005

Fig 6. Phylogentic position of Icaronycteris gunnelli and other Eocene bat fossils with respect to extant bat lineages. A) Strict consensus of three most parsimonious trees of 1511 steps resulting from analyses including all taxa; B) Strict consensus of two most parsimonious trees of 1455 steps for analyses excluding Icaronycteris? menui and Icaronycteris sigei. Bootstrap values from 10,000 bootstrap replicates are shown above and to the left of nodes. Fossil taxa are represented by a dagger (†). https://doi.org/10.1371/journal.pone.0283505.g006